ali asghar jaafarzadeh

Soil salinity and salinization of lands, as one of the main problems of agriculture, has paramount importance and can be avoided with proper understanding of its progress. This is the first step in identifying areas of salt and salt mapping in these soils. With the development of remote sensing technology and efficient use of satellite imaging, this study compared the salinity maps produced by a variety of classification image algorithms (Maximum likelihood, Minimum distance and Parallelepiped) by Landsat5 TM satellite data in the East of Khoy. In this study, 269 soil samples were analyzed and the results obtained were implemented on TM image. For initial identification, topography maps, ENVI 4.8 software and satellite image processing were used and geometric correction was performed with specific locations using GPS. Selected train samples with good distribution in the image and salinity classes were prepared from 1 to 9 determined. Examples of each class of salinity were placed carefully with single pixel size in each image on the corresponding pixel and stored with ROI format, because they had the coordinates. The results indicate the existence of a correlation between bands 1, 4 and 5 TMimage with salinity data and classification algorithms using the Pixel-based method, the highest accuracy of the map is the maximum lkelihood. For this purpose, indicators such as error matrix, Producer’s Accuracy, User’s Accuracy, overall accuracy and kappa Coefficient were extracted. This map is also consistent with field observations of different classes of soil salinity measurement and shows high accuracy of the algorithm among soil salinity maps. The aim of this study was to compare salinity maps prepared by these methods in the area with results of other researchers.